Use case · clinics & medical practices

The claim reconciliation where every rejection ties back to the file

The agent reads the files your clinic already receives — insurer remittance advices, claim files, billing reports — and writes a reconciliation where every payment and rejection points to its source line. Sensitive data only enters the answer for someone allowed to see it.

The question this use case answers

How does a Singapore clinic reconcile insurer and MediSave claims and compute billing per doctor with every figure traceable to its source claim?

The files you already have

  • Insurer / MediShield remittance advices
  • Claim submission files (XML / CSV)
  • Billing reports (.xlsx / CSV)
  • Bank statements (CSV / MT940)
sample

How much did the insurer reject in May?

S$12,480

remittance_may_insurerX.xml · batch 084 · you have access

Illustrative figure (sample) with fictional data — not a customer result or benchmark.

How the agent handles it, step by step

  1. 1

    Upload the month's remittances

    Drop in the files each insurer sends. No integration — the agent identifies the claim files and billing reports by format.

  2. 2

    The agent reconciles claim by claim

    Every payment and rejection carries where it came from: file, batch and line. The expected-versus-paid maths is deterministic, not a model's guess.

  3. 3

    A rejection with no advice becomes a gap

    If an insurer's remittance hasn't arrived, the agent flags the gap instead of closing the reconciliation over an invisible hole.

  4. 4

    Schedule the monthly reconciliation

    Save the recipe and the agent re-runs it each payment cycle, with a versioned audit trail.

Permission at synthesis time

In a clinic, patient data is sensitive data. Redijo checks the permission of each fragment at synthesis time: admin staff see the payment figure without restricted clinical data leaking — not even inside an aggregated total. Permission holds the moment the answer is written, not just at retrieval.

A clinic loses money two ways: the claim rejection nobody contested and the payment nobody checked. Both have the same root — nobody can trace the figure back to the claim. Generic AI tools make it worse: they return a plausible billing summary without saying where each figure came from.

Redijo starts with the proof. The agent reads the insurer remittances, claim files and billing reports your clinic already receives, and writes a reconciliation where every payment and rejection points to its source line.

Auditable by construction, not on trust

The proof is born with the answer. Before writing a figure, the agent locates it in a file you uploaded and records its origin — file, batch, line. The expected-versus-paid comparison is deterministic; the AI writes the prose, never the number. That is why the reconciliation is auditable, not just convincing.

Sensitive data only enters the answer for someone allowed to see it

The second win is privacy. Almost every tool checks access only at retrieval — then lets the AI blend everything. Redijo checks the permission of each fragment at synthesis time. Admin staff close the payment without restricted clinical data appearing, not even hidden inside a total.

Who it’s for

For the clinic manager who needs proof to contest rejections, the partner computing billing per doctor, and whoever answers for PDPA compliance. Your data is hosted in the cloud, with international-transfer safeguards.

Questions this page answers

Your clinic's data is hosted in the cloud, with international-transfer safeguards. PDPA-aligned by architecture — never by local servers.

Reconcile claims with figures you can audit

Get started and close an auditable reconciliation from the files your clinic already receives.

PDPA-ready · DPA for businesses · cancel anytime

← Back to use cases by sector